pack-check.c

   1 #include "cache.h"
   2 #include "hex.h"
   3 #include "repository.h"
   4 #include "pack.h"
   5 #include "pack-revindex.h"
   6 #include "progress.h"
   7 #include "packfile.h"
   8 #include "object-store.h"
   9
  10 struct idx_entry {
  11         off_t                offset;
  12         unsigned int nr;
  13 };
  14
  15 static int compare_entries(const void *e1, const void *e2)
  16 {
  17         const struct idx_entry *entry1 = e1;
  18         const struct idx_entry *entry2 = e2;
  19         if (entry1->offset < entry2->offset)
  20                 return -1;
  21         if (entry1->offset > entry2->offset)
  22                 return 1;
  23         return 0;
  24 }
  25
  26 int check_pack_crc(struct packed_git *p, struct pack_window **w_curs,
  27                    off_t offset, off_t len, unsigned int nr)
  28 {
  29         const uint32_t *index_crc;
  30         uint32_t data_crc = crc32(0, NULL, 0);
  31
  32         do {
  33                 unsigned long avail;
  34                 void *data = use_pack(p, w_curs, offset, &avail);
  35                 if (avail > len)
  36                         avail = len;
  37                 data_crc = crc32(data_crc, data, avail);
  38                 offset += avail;
  39                 len -= avail;
  40         } while (len);
  41
  42         index_crc = p->index_data;
  43         index_crc += 2 + 256 + (size_t)p->num_objects * (the_hash_algo->rawsz/4) + nr;
  44
  45         return data_crc != ntohl(*index_crc);
  46 }
  47
  48 static int verify_packfile(struct repository *r,
  49                            struct packed_git *p,
  50                            struct pack_window **w_curs,
  51                            verify_fn fn,
  52                            struct progress *progress, uint32_t base_count)
  53
  54 {
  55         off_t index_size = p->index_size;
  56         const unsigned char *index_base = p->index_data;
  57         git_hash_ctx ctx;
  58         unsigned char hash[GIT_MAX_RAWSZ], *pack_sig;
  59         off_t offset = 0, pack_sig_ofs = 0;
  60         uint32_t nr_objects, i;
  61         int err = 0;
  62         struct idx_entry *entries;
  63
  64         if (!is_pack_valid(p))
  65                 return error("packfile %s cannot be accessed", p->pack_name);
  66
  67         r->hash_algo->init_fn(&ctx);
  68         do {
  69                 unsigned long remaining;
  70                 unsigned char *in = use_pack(p, w_curs, offset, &remaining);
  71                 offset += remaining;
  72                 if (!pack_sig_ofs)
  73                         pack_sig_ofs = p->pack_size - r->hash_algo->rawsz;
  74                 if (offset > pack_sig_ofs)
  75                         remaining -= (unsigned int)(offset - pack_sig_ofs);
  76                 r->hash_algo->update_fn(&ctx, in, remaining);
  77         } while (offset < pack_sig_ofs);
  78         r->hash_algo->final_fn(hash, &ctx);
  79         pack_sig = use_pack(p, w_curs, pack_sig_ofs, NULL);
  80         if (!hasheq(hash, pack_sig))
  81                 err = error("%s pack checksum mismatch",
  82                             p->pack_name);
  83         if (!hasheq(index_base + index_size - r->hash_algo->hexsz, pack_sig))
  84                 err = error("%s pack checksum does not match its index",
  85                             p->pack_name);
  86         unuse_pack(w_curs);
  87
  88         /* Make sure everything reachable from idx is valid.  Since we
  89          * have verified that nr_objects matches between idx and pack,
  90          * we do not do scan-streaming check on the pack file.
  91          */
  92         nr_objects = p->num_objects;
  93         ALLOC_ARRAY(entries, nr_objects + 1);
  94         entries[nr_objects].offset = pack_sig_ofs;
  95         /* first sort entries by pack offset, since unpacking them is more efficient that way */
  96         for (i = 0; i < nr_objects; i++) {
  97                 entries[i].offset = nth_packed_object_offset(p, i);
  98                 entries[i].nr = i;
  99         }
 100         QSORT(entries, nr_objects, compare_entries);
 101
 102         for (i = 0; i < nr_objects; i++) {
 103                 void *data;
 104                 struct object_id oid;
 105                 enum object_type type;
 106                 unsigned long size;
 107                 off_t curpos;
 108                 int data_valid;
 109
 110                 if (nth_packed_object_id(&oid, p, entries[i].nr) < 0)
 111                         BUG("unable to get oid of object %lu from %s",
 112                             (unsigned long)entries[i].nr, p->pack_name);
 113
 114                 if (p->index_version > 1) {
 115                         off_t offset = entries[i].offset;
 116                         off_t len = entries[i+1].offset - offset;
 117                         unsigned int nr = entries[i].nr;
 118                         if (check_pack_crc(p, w_curs, offset, len, nr))
 119                                 err = error("index CRC mismatch for object %s "
 120                                             "from %s at offset %"PRIuMAX"",
 121                                             oid_to_hex(&oid),
 122                                             p->pack_name, (uintmax_t)offset);
 123                 }
 124
 125                 curpos = entries[i].offset;
 126                 type = unpack_object_header(p, w_curs, &curpos, &size);
 127                 unuse_pack(w_curs);
 128
 129                 if (type == OBJ_BLOB && big_file_threshold <= size) {
 130                         /*
 131                          * Let stream_object_signature() check it with
 132                          * the streaming interface; no point slurping
 133                          * the data in-core only to discard.
 134                          */
 135                         data = NULL;
 136                         data_valid = 0;
 137                 } else {
 138                         data = unpack_entry(r, p, entries[i].offset, &type, &size);
 139                         data_valid = 1;
 140                 }
 141
 142                 if (data_valid && !data)
 143                         err = error("cannot unpack %s from %s at offset %"PRIuMAX"",
 144                                     oid_to_hex(&oid), p->pack_name,
 145                                     (uintmax_t)entries[i].offset);
 146                 else if (data && check_object_signature(r, &oid, data, size,
 147                                                         type) < 0)
 148                         err = error("packed %s from %s is corrupt",
 149                                     oid_to_hex(&oid), p->pack_name);
 150                 else if (!data && stream_object_signature(r, &oid) < 0)
 151                         err = error("packed %s from %s is corrupt",
 152                                     oid_to_hex(&oid), p->pack_name);
 153                 else if (fn) {
 154                         int eaten = 0;
 155                         err |= fn(&oid, type, size, data, &eaten);
 156                         if (eaten)
 157                                 data = NULL;
 158                 }
 159                 if (((base_count + i) & 1023) == 0)
 160                         display_progress(progress, base_count + i);
 161                 free(data);
 162
 163         }
 164         display_progress(progress, base_count + i);
 165         free(entries);
 166
 167         return err;
 168 }
 169
 170 int verify_pack_index(struct packed_git *p)
 171 {
 172         int err = 0;
 173
 174         if (open_pack_index(p))
 175                 return error("packfile %s index not opened", p->pack_name);
 176
 177         /* Verify SHA1 sum of the index file */
 178         if (!hashfile_checksum_valid(p->index_data, p->index_size))
 179                 err = error("Packfile index for %s hash mismatch",
 180                             p->pack_name);
 181         return err;
 182 }
 183
 184 int verify_pack(struct repository *r, struct packed_git *p, verify_fn fn,
 185                 struct progress *progress, uint32_t base_count)
 186 {
 187         int err = 0;
 188         struct pack_window *w_curs = NULL;
 189
 190         err |= verify_pack_index(p);
 191         if (!p->index_data)
 192                 return -1;
 193
 194         err |= verify_packfile(r, p, &w_curs, fn, progress, base_count);
 195         unuse_pack(&w_curs);
 196
 197         return err;
 198 }